Playoff Odds
20,000 simulations of the 256 games still to play. Results already on the board are kept as results, so these are odds given the season so far rather than a fresh guess each week.
| Team | Div | Rec | Proj W | Playoffs ▾ | Division | 1 seed |
|---|---|---|---|---|---|---|
 JAX | South | 1-0 | 10.5 | 76% | 62% | 16% |
 BAL | North | 1-0 | 10.6 | 74% | 45% | 17% |
 KC | West | 1-0 | 10.3 | 71% | 49% | 15% |
 NYJ | East | 1-0 | 9.8 | 63% | 41% | 10% |
 LV | West | 1-0 | 9.5 | 60% | 36% | 9% |
 BUF | East | 1-0 | 9.4 | 57% | 35% | 8% |
 CIN | North | 1-0 | 9.4 | 57% | 27% | 9% |
 PIT | North | 1-0 | 9.2 | 53% | 23% | 7% |
 HOU | South | 0-1 | 7.7 | 32% | 17% | 2% |
 MIA | East | 0-1 | 7.4 | 28% | 12% | 2% |
 NE | East | 0-1 | 7.4 | 28% | 12% | 2% |
 TEN | South | 0-1 | 7.2 | 26% | 13% | 1% |
 LAC | West | 0-1 | 6.9 | 21% | 9% | 1% |
 DEN | West | 0-1 | 6.6 | 18% | 7% | 1% |
 CLE | North | 0-1 | 6.6 | 18% | 5% | 1% |
 IND | South | 0-1 | 6.5 | 18% | 8% | 1% |
How this is worked out. Each team carries a rating. Net EPA per play turned into a point spread, and every remaining game is decided by that gap plus two points for playing at home, through a logistic curve scaled to NFL single-game uncertainty.
Every run also draws each team a season-long rating error of about ±5.8 points. Without it the model had a team at 100% to reach the playoffs after one week, which is not a forecast so much as a refusal to admit that last season’s numbers are a prior rather than the truth. That band narrows as real games accrue, and for the same reason, ratings themselves are shrunk toward the league average early on, because one week of EPA spread the league across 71 points when the real gap between best and worst is nearer 20.
Ties for seeding are broken at random. The real chain is head-to-head, then division record, then common games, then conference record, and this model tracks none of it. Random breaks are unbiased across twenty thousand runs, but they are not a claim about any particular tie.